COMPARATIVE ANALYSIS OF DIVERSE MACHINE LEARNING TECHNIQUES TO DETECT CYBERBULLYING IN FACEBOOK

Since the utilization of social media networks have advanced, the capacity of its end users to torment others has raised up too. One of the pervasive types of tormenting is Cyberbullying, which happens on the social media networks like Facebook, Twitter, Instagram, etc. The previous decennium has seen a development in cyberbullying as a type of harassing that happens basically with the utilization of electronic gadgets, for instance, social media, email, internet gaming or via pictures or unwanted context sent to those devices. This study proposes a methodology for detecting cyberbullying from the comments of Facebook, the widely used social media in current trend. In order to investigate the issue, a newly created dataset with the utilization Facebook Graph API explorer has been incorporated. Total of three ML classification algorithms were utilized, namely, Random Forest, Support Vector Machine and Naïve Bayes along with Term Frequency-Inverse Document Frequency (TF-IDF) Feature extraction strategy. Every one of these classifiers was assessed by utilizing widely used metrics such as Accuracy, Precision, Recall and F1-Score to decide the ML classifiers’ rate of detecting applied to the created dataset.